def visualize(df, data, n_data, dataset_name):
visualize_scores(
df,
score_names=["rmse"],
is_higher_score_better=[False],
# err_param_name="std",
# err_param_name="magnitude",
err_param_name="prob_break",
title=
f"Prediction scores for {dataset_name} dataset (n={n_data}) with added error"
)
visualize_time_series_prediction(
df,
data,
score_name="rmse",
is_higher_score_better=False,
# err_param_name="std",
# err_param_name="magnitude",
err_param_name="prob_break",
model_name="LSTM",
err_train_column="err_train",
test_pred_column="test_pred",
title=
f"Predictions for {dataset_name} dataset (n={n_data}) with added error"
)
plt.show()
def visualize(df, dataset_name, label_names, test_data, use_interactive_mode):
visualize_scores(
df,
score_names=["test_mean_accuracy", "train_mean_accuracy"],
is_higher_score_better=[True, True],
err_param_name="p",
title=f"{dataset_name} classification scores with added error"
)
visualize_best_model_params(
df,
"MultinomialNB",
model_params=["alpha"],
score_names=["test_mean_accuracy"],
is_higher_score_better=[True],
err_param_name="p",
title=f"Best parameters for {dataset_name} classification",
y_log=True
)
visualize_best_model_params(
df,
"LinearSVC",
model_params=["C"],
score_names=["test_mean_accuracy"],
is_higher_score_better=[True],
err_param_name="p",
title=f"Best parameters for {dataset_name} classification",
y_log=True
)
visualize_classes(
df,
label_names,
err_param_name="p",
reduced_data_column="reduced_test_data",
labels_column="test_labels",
cmap="tab20",
title=f"{dataset_name} test set (n={len(test_data)}) true classes with added error"
)
if use_interactive_mode:
def on_click(element, label, predicted_label):
print(label, " predicted as ", predicted_label, ":", sep="")
print(element, end="\n\n")
else:
on_click = None
visualize_confusion_matrices(
df,
label_names,
score_name="test_mean_accuracy",
is_higher_score_better=True,
err_param_name="p",
labels_col="test_labels",
predictions_col="predicted_test_labels",
on_click=on_click
)
plt.show()
def visualize(df):
# visualize_scores(df, ["mAP-50"], [True], "std", "Object detection with Gaussian noise", x_log=False)
# visualize_scores(df, ["mAP-50"], [True], "std", "Object detection with Gaussian blur", x_log=False)
# visualize_scores(df, ["mAP-50"], [True], "snowflake_probability", "Object detection with snow filter", x_log=True)
# visualize_scores(df, ["mAP-50"], [True], "probability", "Object detection with rain filter", x_log=True)
# visualize_scores(df, ["mAP-50"], [True], "probability", "Object detection with added stains", x_log=True)
# visualize_scores(df, ["mAP-50"], [True], "quality", "Object detection with JPEG compression", x_log=False)
visualize_scores(df, ["mAP-50"], [True], "k", "Object detection with reduced resolution", x_log=False)
# visualize_scores(df, ["mAP-50"], [True], "rat", "Object detection with added brightness", x_log=False)
plt.show()
def visualize(df, label_names, dataset_name, data, use_interactive_mode):
visualize_scores(
df,
score_names=["AMI", "ARI"],
is_higher_score_better=[True, True],
# err_param_name="std",
# err_param_name="probability",
err_param_name="max_angle",
# title=f"{dataset_name} clustering scores with added gaussian noise",
# title=f"{dataset_name} clustering scores with missing pixels",
title=f"{dataset_name} clustering scores with rotation",
)
visualize_best_model_params(
df,
model_name="HDBSCAN",
model_params=["min_cluster_size", "min_samples"],
score_names=["AMI", "ARI"],
is_higher_score_better=[True, True],
# err_param_name="std",
# err_param_name="probability",
err_param_name="max_angle",
title=f"Best parameters for {dataset_name} clustering")
visualize_classes(
df,
label_names,
# err_param_name="std",
# err_param_name="probability",
err_param_name="max_angle",
reduced_data_column="reduced_data",
labels_column="labels",
cmap="tab10",
# title=f"{dataset_name} (n={data.shape[0]}) classes with added gaussian noise"
# title=f"{dataset_name} (n={data.shape[0]}) classes with missing pixels"
title=f"{dataset_name} (n={data.shape[0]}) classes with rotation")
if use_interactive_mode:
def on_click(original, modified):
# reshape data
original = original.reshape((28, 28))
modified = modified.reshape((28, 28))
# create a figure and draw the images
fg, axs = plt.subplots(1, 2)
axs[0].matshow(original, cmap='gray_r')
axs[0].axis('off')
axs[1].matshow(modified, cmap='gray_r')
axs[1].axis('off')
fg.show()
# Remember to enable runner's interactive mode
visualize_interactive_plot(df, "max_angle", data, "tab10",
"reduced_data", on_click)
plt.show()
def visualize(df):
# Visualize mean squared error for all used standard deviations
visualize_scores(df=df,
score_names=["MSE"],
is_higher_score_better=[False],
err_param_name="std",
title="Mean squared error")
visualize_best_model_params(df=df,
model_name="Predictor #1",
model_params=["weight"],
score_names=["MSE"],
is_higher_score_better=[False],
err_param_name="std",
title=f"Best model params")
plt.show()